chore: add host resources related graphs

[fryorcraken]

Would be good to add to the default Grafana some host resources monitoring to better gauge resource usage and whether upgrade is needed:

• memory usage
• io usage
• CPU usage

Some less useful but WoW effect metrics could also be added such as:

• location of nodes connected to.

[vpavlin]

We might need to run additional container for the resource metrics (cAdvisor) - @alrevuelta already has it in the waku-simulator, so it would be easy to integrate

For the location, I don't think it would be that easy - we have this in network-monitor, but the way it works is that NM calls some IP-to-location service and then exposes that vie metrics. So we could have it, but we would have to figure out how/where to translate the peer IP to location (and I don't think it belongs to nwaku, so maybe an external tool querying nwaku REST for peers info and then doing the translation?)

[alrevuelta]

mem/cpu metrics, sure, but I would leave the location out of scope.

Note though that at some point we might need to split nwaku-compose into two: "professional operators" and "average users". We already have prometheus, grafana, postgres + (cadvisor for this?) services, so this setup starts looking more like a professional node operator one than for a user that just wants to quickly run a nwaku node. Unsure though where nwaku-compose should sit.

[vpavlin]

We can always provide more guidance on starting particular services, but I think as long as the default behaviour of docker-compose up [-d] just works and exposed ports do not colide, it does not really matter how much is started